includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         var $list_authors = false; # Return distinct author list (when not returning full history)
  35         var $author_list = "";
  36
  37         var $dumpUploads = false;
  38         var $dumpUploadFileContents = false;
  39
  40         const FULL = 1;
  41         const CURRENT = 2;
  42         const STABLE = 4; // extension defined
  43         const LOGS = 8;
  44         const RANGE = 16;
  45
  46         const BUFFER = 0;
  47         const STREAM = 1;
  48
  49         const TEXT = 0;
  50         const STUB = 1;
  51
  52         var $buffer;
  53
  54         var $text;
  55
  56         /**
  57          * @var DumpOutput
  58          */
  59         var $sink;
  60
  61         /**
  62          * Returns the export schema version.
  63          * @return string
  64          */
  65         public static function schemaVersion() {
  66                 return "0.8";
  67         }
  68
  69         /**
  70          * If using WikiExporter::STREAM to stream a large amount of data,
  71          * provide a database connection which is not managed by
  72          * LoadBalancer to read from: some history blob types will
  73          * make additional queries to pull source data while the
  74          * main query is still running.
  75          *
  76          * @param $db DatabaseBase
  77          * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
  78          *                 WikiExporter::RANGE or WikiExporter::STABLE,
  79          *                 or an associative array:
  80          *                   offset: non-inclusive offset at which to start the query
  81          *                   limit: maximum number of rows to return
  82          *                   dir: "asc" or "desc" timestamp order
  83          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  84          * @param int $text one of WikiExporter::TEXT or WikiExporter::STUB
  85          */
  86         function __construct( $db, $history = WikiExporter::CURRENT,
  87                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  88                 $this->db = $db;
  89                 $this->history = $history;
  90                 $this->buffer = $buffer;
  91                 $this->writer = new XmlDumpWriter();
  92                 $this->sink = new DumpOutput();
  93                 $this->text = $text;
  94         }
  95
  96         /**
  97          * Set the DumpOutput or DumpFilter object which will receive
  98          * various row objects and XML output for filtering. Filters
  99          * can be chained or used as callbacks.
 100          *
 101          * @param $sink mixed
 102          */
 103         public function setOutputSink( &$sink ) {
 104                 $this->sink =& $sink;
 105         }
 106
 107         public function openStream() {
 108                 $output = $this->writer->openStream();
 109                 $this->sink->writeOpenStream( $output );
 110         }
 111
 112         public function closeStream() {
 113                 $output = $this->writer->closeStream();
 114                 $this->sink->writeCloseStream( $output );
 115         }
 116
 117         /**
 118          * Dumps a series of page and revision records for all pages
 119          * in the database, either including complete history or only
 120          * the most recent version.
 121          */
 122         public function allPages() {
 123                 $this->dumpFrom( '' );
 124         }
 125
 126         /**
 127          * Dumps a series of page and revision records for those pages
 128          * in the database falling within the page_id range given.
 129          * @param int $start inclusive lower limit (this id is included)
 130          * @param $end   Int: Exclusive upper limit (this id is not included)
 131          *                   If 0, no upper limit.
 132          */
 133         public function pagesByRange( $start, $end ) {
 134                 $condition = 'page_id >= ' . intval( $start );
 135                 if ( $end ) {
 136                         $condition .= ' AND page_id < ' . intval( $end );
 137                 }
 138                 $this->dumpFrom( $condition );
 139         }
 140
 141         /**
 142          * Dumps a series of page and revision records for those pages
 143          * in the database with revisions falling within the rev_id range given.
 144          * @param int $start inclusive lower limit (this id is included)
 145          * @param $end   Int: Exclusive upper limit (this id is not included)
 146          *                   If 0, no upper limit.
 147          */
 148         public function revsByRange( $start, $end ) {
 149                 $condition = 'rev_id >= ' . intval( $start );
 150                 if ( $end ) {
 151                         $condition .= ' AND rev_id < ' . intval( $end );
 152                 }
 153                 $this->dumpFrom( $condition );
 154         }
 155
 156         /**
 157          * @param $title Title
 158          */
 159         public function pageByTitle( $title ) {
 160                 $this->dumpFrom(
 161                         'page_namespace=' . $title->getNamespace() .
 162                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 163         }
 164
 165         /**
 166          * @param $name string
 167          * @throws MWException
 168          */
 169         public function pageByName( $name ) {
 170                 $title = Title::newFromText( $name );
 171                 if ( is_null( $title ) ) {
 172                         throw new MWException( "Can't export invalid title" );
 173                 } else {
 174                         $this->pageByTitle( $title );
 175                 }
 176         }
 177
 178         /**
 179          * @param $names array
 180          */
 181         public function pagesByName( $names ) {
 182                 foreach ( $names as $name ) {
 183                         $this->pageByName( $name );
 184                 }
 185         }
 186
 187         public function allLogs() {
 188                 $this->dumpFrom( '' );
 189         }
 190
 191         /**
 192          * @param $start int
 193          * @param $end int
 194          */
 195         public function logsByRange( $start, $end ) {
 196                 $condition = 'log_id >= ' . intval( $start );
 197                 if ( $end ) {
 198                         $condition .= ' AND log_id < ' . intval( $end );
 199                 }
 200                 $this->dumpFrom( $condition );
 201         }
 202
 203         /**
 204          * Generates the distinct list of authors of an article
 205          * Not called by default (depends on $this->list_authors)
 206          * Can be set by Special:Export when not exporting whole history
 207          *
 208          * @param $cond
 209          */
 210         protected function do_list_authors( $cond ) {
 211                 wfProfileIn( __METHOD__ );
 212                 $this->author_list = "<contributors>";
 213                 // rev_deleted
 214
 215                 $res = $this->db->select(
 216                         array( 'page', 'revision' ),
 217                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 218                         array(
 219                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 220                                 $cond,
 221                                 'page_id = rev_id',
 222                         ),
 223                         __METHOD__
 224                 );
 225
 226                 foreach ( $res as $row ) {
 227                         $this->author_list .= "<contributor>" .
 228                                 "<username>" .
 229                                 htmlentities( $row->rev_user_text ) .
 230                                 "</username>" .
 231                                 "<id>" .
 232                                 $row->rev_user .
 233                                 "</id>" .
 234                                 "</contributor>";
 235                 }
 236                 $this->author_list .= "</contributors>";
 237                 wfProfileOut( __METHOD__ );
 238         }
 239
 240         /**
 241          * @param $cond string
 242          * @throws MWException
 243          * @throws Exception
 244          */
 245         protected function dumpFrom( $cond = '' ) {
 246                 wfProfileIn( __METHOD__ );
 247                 # For logging dumps...
 248                 if ( $this->history & self::LOGS ) {
 249                         $where = array( 'user_id = log_user' );
 250                         # Hide private logs
 251                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 252                         if ( $hideLogs ) {
 253                                 $where[] = $hideLogs;
 254                         }
 255                         # Add on any caller specified conditions
 256                         if ( $cond ) {
 257                                 $where[] = $cond;
 258                         }
 259                         # Get logging table name for logging.* clause
 260                         $logging = $this->db->tableName( 'logging' );
 261
 262                         if ( $this->buffer == WikiExporter::STREAM ) {
 263                                 $prev = $this->db->bufferResults( false );
 264                         }
 265                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 266                         try {
 267                                 $result = $this->db->select( array( 'logging', 'user' ),
 268                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 269                                         $where,
 270                                         __METHOD__,
 271                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 272                                 );
 273                                 $wrapper = $this->db->resultObject( $result );
 274                                 $this->outputLogStream( $wrapper );
 275                                 if ( $this->buffer == WikiExporter::STREAM ) {
 276                                         $this->db->bufferResults( $prev );
 277                                 }
 278                         } catch ( Exception $e ) {
 279                                 // Throwing the exception does not reliably free the resultset, and
 280                                 // would also leave the connection in unbuffered mode.
 281
 282                                 // Freeing result
 283                                 try {
 284                                         if ( $wrapper ) {
 285                                                 $wrapper->free();
 286                                         }
 287                                 } catch ( Exception $e2 ) {
 288                                         // Already in panic mode -> ignoring $e2 as $e has
 289                                         // higher priority
 290                                 }
 291
 292                                 // Putting database back in previous buffer mode
 293                                 try {
 294                                         if ( $this->buffer == WikiExporter::STREAM ) {
 295                                                 $this->db->bufferResults( $prev );
 296                                         }
 297                                 } catch ( Exception $e2 ) {
 298                                         // Already in panic mode -> ignoring $e2 as $e has
 299                                         // higher priority
 300                                 }
 301
 302                                 // Inform caller about problem
 303                                 wfProfileOut( __METHOD__ );
 304                                 throw $e;
 305                         }
 306                 # For page dumps...
 307                 } else {
 308                         $tables = array( 'page', 'revision' );
 309                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 310                         $opts['USE INDEX'] = array();
 311                         $join = array();
 312                         if ( is_array( $this->history ) ) {
 313                                 # Time offset/limit for all pages/history...
 314                                 $revJoin = 'page_id=rev_page';
 315                                 # Set time order
 316                                 if ( $this->history['dir'] == 'asc' ) {
 317                                         $op = '>';
 318                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 319                                 } else {
 320                                         $op = '<';
 321                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 322                                 }
 323                                 # Set offset
 324                                 if ( !empty( $this->history['offset'] ) ) {
 325                                         $revJoin .= " AND rev_timestamp $op " .
 326                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 327                                 }
 328                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 329                                 # Set query limit
 330                                 if ( !empty( $this->history['limit'] ) ) {
 331                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 332                                 }
 333                         } elseif ( $this->history & WikiExporter::FULL ) {
 334                                 # Full history dumps...
 335                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 336                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 337                                 # Latest revision dumps...
 338                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 339                                         $this->do_list_authors( $cond );
 340                                 }
 341                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 342                         } elseif ( $this->history & WikiExporter::STABLE ) {
 343                                 # "Stable" revision dumps...
 344                                 # Default JOIN, to be overridden...
 345                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 346                                 # One, and only one hook should set this, and return false
 347                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 348                                         wfProfileOut( __METHOD__ );
 349                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 350                                 }
 351                         } elseif ( $this->history & WikiExporter::RANGE ) {
 352                                 # Dump of revisions within a specified range
 353                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 354                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 355                         } else {
 356                                 # Unknown history specification parameter?
 357                                 wfProfileOut( __METHOD__ );
 358                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 359                         }
 360                         # Query optimization hacks
 361                         if ( $cond == '' ) {
 362                                 $opts[] = 'STRAIGHT_JOIN';
 363                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 364                         }
 365                         # Build text join options
 366                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 367                                 $tables[] = 'text';
 368                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 369                         }
 370
 371                         if ( $this->buffer == WikiExporter::STREAM ) {
 372                                 $prev = $this->db->bufferResults( false );
 373                         }
 374
 375                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 376                         try {
 377                                 wfRunHooks( 'ModifyExportQuery',
 378                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 379
 380                                 # Do the query!
 381                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 382                                 $wrapper = $this->db->resultObject( $result );
 383                                 # Output dump results
 384                                 $this->outputPageStream( $wrapper );
 385
 386                                 if ( $this->buffer == WikiExporter::STREAM ) {
 387                                         $this->db->bufferResults( $prev );
 388                                 }
 389                         } catch ( Exception $e ) {
 390                                 // Throwing the exception does not reliably free the resultset, and
 391                                 // would also leave the connection in unbuffered mode.
 392
 393                                 // Freeing result
 394                                 try {
 395                                         if ( $wrapper ) {
 396                                                 $wrapper->free();
 397                                         }
 398                                 } catch ( Exception $e2 ) {
 399                                         // Already in panic mode -> ignoring $e2 as $e has
 400                                         // higher priority
 401                                 }
 402
 403                                 // Putting database back in previous buffer mode
 404                                 try {
 405                                         if ( $this->buffer == WikiExporter::STREAM ) {
 406                                                 $this->db->bufferResults( $prev );
 407                                         }
 408                                 } catch ( Exception $e2 ) {
 409                                         // Already in panic mode -> ignoring $e2 as $e has
 410                                         // higher priority
 411                                 }
 412
 413                                 // Inform caller about problem
 414                                 throw $e;
 415                         }
 416                 }
 417                 wfProfileOut( __METHOD__ );
 418         }
 419
 420         /**
 421          * Runs through a query result set dumping page and revision records.
 422          * The result set should be sorted/grouped by page to avoid duplicate
 423          * page records in the output.
 424          *
 425          * Should be safe for
 426          * streaming (non-buffered) queries, as long as it was made on a
 427          * separate database connection not managed by LoadBalancer; some
 428          * blob storage types will make queries to pull source data.
 429          *
 430          * @param $resultset ResultWrapper
 431          */
 432         protected function outputPageStream( $resultset ) {
 433                 $last = null;
 434                 foreach ( $resultset as $row ) {
 435                         if ( $last === null ||
 436                                 $last->page_namespace != $row->page_namespace ||
 437                                 $last->page_title != $row->page_title ) {
 438                                 if ( $last !== null ) {
 439                                         $output = '';
 440                                         if ( $this->dumpUploads ) {
 441                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 442                                         }
 443                                         $output .= $this->writer->closePage();
 444                                         $this->sink->writeClosePage( $output );
 445                                 }
 446                                 $output = $this->writer->openPage( $row );
 447                                 $this->sink->writeOpenPage( $row, $output );
 448                                 $last = $row;
 449                         }
 450                         $output = $this->writer->writeRevision( $row );
 451                         $this->sink->writeRevision( $row, $output );
 452                 }
 453                 if ( $last !== null ) {
 454                         $output = '';
 455                         if ( $this->dumpUploads ) {
 456                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 457                         }
 458                         $output .= $this->author_list;
 459                         $output .= $this->writer->closePage();
 460                         $this->sink->writeClosePage( $output );
 461                 }
 462         }
 463
 464         /**
 465          * @param $resultset array
 466          */
 467         protected function outputLogStream( $resultset ) {
 468                 foreach ( $resultset as $row ) {
 469                         $output = $this->writer->writeLogItem( $row );
 470                         $this->sink->writeLogItem( $row, $output );
 471                 }
 472         }
 473 }
 474
 475 /**
 476  * @ingroup Dump
 477  */
 478 class XmlDumpWriter {
 479         /**
 480          * Returns the export schema version.
 481          * @deprecated in 1.20; use WikiExporter::schemaVersion() instead
 482          * @return string
 483          */
 484         function schemaVersion() {
 485                 wfDeprecated( __METHOD__, '1.20' );
 486                 return WikiExporter::schemaVersion();
 487         }
 488
 489         /**
 490          * Opens the XML output stream's root "<mediawiki>" element.
 491          * This does not include an xml directive, so is safe to include
 492          * as a subelement in a larger XML stream. Namespace and XML Schema
 493          * references are included.
 494          *
 495          * Output will be encoded in UTF-8.
 496          *
 497          * @return string
 498          */
 499         function openStream() {
 500                 global $wgLanguageCode;
 501                 $ver = WikiExporter::schemaVersion();
 502                 return Xml::element( 'mediawiki', array(
 503                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 504                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 505                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 506                                 #TODO: how do we get a new version up there?
 507                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 508                         'version'            => $ver,
 509                         'xml:lang'           => $wgLanguageCode ),
 510                         null ) .
 511                         "\n" .
 512                         $this->siteInfo();
 513         }
 514
 515         /**
 516          * @return string
 517          */
 518         function siteInfo() {
 519                 $info = array(
 520                         $this->sitename(),
 521                         $this->homelink(),
 522                         $this->generator(),
 523                         $this->caseSetting(),
 524                         $this->namespaces() );
 525                 return "  <siteinfo>\n    " .
 526                         implode( "\n    ", $info ) .
 527                         "\n  </siteinfo>\n";
 528         }
 529
 530         /**
 531          * @return string
 532          */
 533         function sitename() {
 534                 global $wgSitename;
 535                 return Xml::element( 'sitename', array(), $wgSitename );
 536         }
 537
 538         /**
 539          * @return string
 540          */
 541         function generator() {
 542                 global $wgVersion;
 543                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 544         }
 545
 546         /**
 547          * @return string
 548          */
 549         function homelink() {
 550                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 551         }
 552
 553         /**
 554          * @return string
 555          */
 556         function caseSetting() {
 557                 global $wgCapitalLinks;
 558                 // "case-insensitive" option is reserved for future
 559                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 560                 return Xml::element( 'case', array(), $sensitivity );
 561         }
 562
 563         /**
 564          * @return string
 565          */
 566         function namespaces() {
 567                 global $wgContLang;
 568                 $spaces = "<namespaces>\n";
 569                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 570                         $spaces .= '      ' .
 571                                 Xml::element( 'namespace',
 572                                         array(
 573                                                 'key' => $ns,
 574                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 575                                         ), $title ) . "\n";
 576                 }
 577                 $spaces .= "    </namespaces>";
 578                 return $spaces;
 579         }
 580
 581         /**
 582          * Closes the output stream with the closing root element.
 583          * Call when finished dumping things.
 584          *
 585          * @return string
 586          */
 587         function closeStream() {
 588                 return "</mediawiki>\n";
 589         }
 590
 591         /**
 592          * Opens a "<page>" section on the output stream, with data
 593          * from the given database row.
 594          *
 595          * @param $row object
 596          * @return string
 597          */
 598         public function openPage( $row ) {
 599                 $out = "  <page>\n";
 600                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 601                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 602                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 603                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 604                 if ( $row->page_is_redirect ) {
 605                         $page = WikiPage::factory( $title );
 606                         $redirect = $page->getRedirectTarget();
 607                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 608                                 $out .= '    ';
 609                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 610                                 $out .= "\n";
 611                         }
 612                 }
 613
 614                 if ( $row->page_restrictions != '' ) {
 615                         $out .= '    ' . Xml::element( 'restrictions', array(),
 616                                 strval( $row->page_restrictions ) ) . "\n";
 617                 }
 618
 619                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 620
 621                 return $out;
 622         }
 623
 624         /**
 625          * Closes a "<page>" section on the output stream.
 626          *
 627          * @access private
 628          * @return string
 629          */
 630         function closePage() {
 631                 return "  </page>\n";
 632         }
 633
 634         /**
 635          * Dumps a "<revision>" section on the output stream, with
 636          * data filled in from the given database row.
 637          *
 638          * @param $row object
 639          * @return string
 640          * @access private
 641          */
 642         function writeRevision( $row ) {
 643                 wfProfileIn( __METHOD__ );
 644
 645                 $out = "    <revision>\n";
 646                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 647                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 648                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 649                 }
 650
 651                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 652
 653                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 654                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 655                 } else {
 656                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 657                 }
 658
 659                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 660                         $out .= "      <minor/>\n";
 661                 }
 662                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 663                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 664                 } elseif ( $row->rev_comment != '' ) {
 665                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 666                 }
 667
 668                 $text = '';
 669                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 670                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 671                 } elseif ( isset( $row->old_text ) ) {
 672                         // Raw text from the database may have invalid chars
 673                         $text = strval( Revision::getRevisionText( $row ) );
 674                         $out .= "      " . Xml::elementClean( 'text',
 675                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 676                                 strval( $text ) ) . "\n";
 677                 } else {
 678                         // Stub output
 679                         $out .= "      " . Xml::element( 'text',
 680                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 681                                 "" ) . "\n";
 682                 }
 683
 684                 if ( isset( $row->rev_sha1 )
 685                         && $row->rev_sha1
 686                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 687                 ) {
 688                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 689                 } else {
 690                         $out .= "      <sha1/>\n";
 691                 }
 692
 693                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 694                         $content_model = strval( $row->rev_content_model );
 695                 } else {
 696                         // probably using $wgContentHandlerUseDB = false;
 697                         // @todo test!
 698                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 699                         $content_model = ContentHandler::getDefaultModelFor( $title );
 700                 }
 701
 702                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 703
 704                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 705                         $content_format = strval( $row->rev_content_format );
 706                 } else {
 707                         // probably using $wgContentHandlerUseDB = false;
 708                         // @todo test!
 709                         $content_handler = ContentHandler::getForModelID( $content_model );
 710                         $content_format = $content_handler->getDefaultFormat();
 711                 }
 712
 713                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 714
 715                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 716
 717                 $out .= "    </revision>\n";
 718
 719                 wfProfileOut( __METHOD__ );
 720                 return $out;
 721         }
 722
 723         /**
 724          * Dumps a "<logitem>" section on the output stream, with
 725          * data filled in from the given database row.
 726          *
 727          * @param $row object
 728          * @return string
 729          * @access private
 730          */
 731         function writeLogItem( $row ) {
 732                 wfProfileIn( __METHOD__ );
 733
 734                 $out = "  <logitem>\n";
 735                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 736
 737                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 738
 739                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 740                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 741                 } else {
 742                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 743                 }
 744
 745                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 746                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 747                 } elseif ( $row->log_comment != '' ) {
 748                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 749                 }
 750
 751                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 752                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 753
 754                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 755                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 756                 } else {
 757                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 758                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 759                         $out .= "    " . Xml::elementClean( 'params',
 760                                 array( 'xml:space' => 'preserve' ),
 761                                 strval( $row->log_params ) ) . "\n";
 762                 }
 763
 764                 $out .= "  </logitem>\n";
 765
 766                 wfProfileOut( __METHOD__ );
 767                 return $out;
 768         }
 769
 770         /**
 771          * @param $timestamp string
 772          * @param string $indent Default to six spaces
 773          * @return string
 774          */
 775         function writeTimestamp( $timestamp, $indent = "      " ) {
 776                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 777                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 778         }
 779
 780         /**
 781          * @param $id
 782          * @param $text string
 783          * @param string $indent Default to six spaces
 784          * @return string
 785          */
 786         function writeContributor( $id, $text, $indent = "      " ) {
 787                 $out = $indent . "<contributor>\n";
 788                 if ( $id || !IP::isValid( $text ) ) {
 789                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 790                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 791                 } else {
 792                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 793                 }
 794                 $out .= $indent . "</contributor>\n";
 795                 return $out;
 796         }
 797
 798         /**
 799          * Warning! This data is potentially inconsistent. :(
 800          * @param $row
 801          * @param $dumpContents bool
 802          * @return string
 803          */
 804         function writeUploads( $row, $dumpContents = false ) {
 805                 if ( $row->page_namespace == NS_FILE ) {
 806                         $img = wfLocalFile( $row->page_title );
 807                         if ( $img && $img->exists() ) {
 808                                 $out = '';
 809                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 810                                         $out .= $this->writeUpload( $ver, $dumpContents );
 811                                 }
 812                                 $out .= $this->writeUpload( $img, $dumpContents );
 813                                 return $out;
 814                         }
 815                 }
 816                 return '';
 817         }
 818
 819         /**
 820          * @param File $file
 821          * @param $dumpContents bool
 822          * @return string
 823          */
 824         function writeUpload( $file, $dumpContents = false ) {
 825                 if ( $file->isOld() ) {
 826                         $archiveName = "      " .
 827                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 828                 } else {
 829                         $archiveName = '';
 830                 }
 831                 if ( $dumpContents ) {
 832                         $be = $file->getRepo()->getBackend();
 833                         # Dump file as base64
 834                         # Uses only XML-safe characters, so does not need escaping
 835                         # @todo Too bad this loads the contents into memory (script might swap)
 836                         $contents = '      <contents encoding="base64">' .
 837                                 chunk_split( base64_encode(
 838                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 839                                 "      </contents>\n";
 840                 } else {
 841                         $contents = '';
 842                 }
 843                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 844                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 845                 } else {
 846                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 847                 }
 848                 return "    <upload>\n" .
 849                         $this->writeTimestamp( $file->getTimestamp() ) .
 850                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 851                         "      " . $comment . "\n" .
 852                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 853                         $archiveName .
 854                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 855                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 856                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 857                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 858                         $contents .
 859                         "    </upload>\n";
 860         }
 861
 862         /**
 863          * Return prefixed text form of title, but using the content language's
 864          * canonical namespace. This skips any special-casing such as gendered
 865          * user namespaces -- which while useful, are not yet listed in the
 866          * XML "<siteinfo>" data so are unsafe in export.
 867          *
 868          * @param Title $title
 869          * @return string
 870          * @since 1.18
 871          */
 872         public static function canonicalTitle( Title $title ) {
 873                 if ( $title->isExternal() ) {
 874                         return $title->getPrefixedText();
 875                 }
 876
 877                 global $wgContLang;
 878                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 879
 880                 if ( $prefix !== '' ) {
 881                         $prefix .= ':';
 882                 }
 883
 884                 return $prefix . $title->getText();
 885         }
 886 }
 887
 888 /**
 889  * Base class for output stream; prints to stdout or buffer or wherever.
 890  * @ingroup Dump
 891  */
 892 class DumpOutput {
 893
 894         /**
 895          * @param $string string
 896          */
 897         function writeOpenStream( $string ) {
 898                 $this->write( $string );
 899         }
 900
 901         /**
 902          * @param $string string
 903          */
 904         function writeCloseStream( $string ) {
 905                 $this->write( $string );
 906         }
 907
 908         /**
 909          * @param $page
 910          * @param $string string
 911          */
 912         function writeOpenPage( $page, $string ) {
 913                 $this->write( $string );
 914         }
 915
 916         /**
 917          * @param $string string
 918          */
 919         function writeClosePage( $string ) {
 920                 $this->write( $string );
 921         }
 922
 923         /**
 924          * @param $rev
 925          * @param $string string
 926          */
 927         function writeRevision( $rev, $string ) {
 928                 $this->write( $string );
 929         }
 930
 931         /**
 932          * @param $rev
 933          * @param $string string
 934          */
 935         function writeLogItem( $rev, $string ) {
 936                 $this->write( $string );
 937         }
 938
 939         /**
 940          * Override to write to a different stream type.
 941          * @param $string string
 942          * @return bool
 943          */
 944         function write( $string ) {
 945                 print $string;
 946         }
 947
 948         /**
 949          * Close the old file, move it to a specified name,
 950          * and reopen new file with the old name. Use this
 951          * for writing out a file in multiple pieces
 952          * at specified checkpoints (e.g. every n hours).
 953          * @param $newname mixed File name. May be a string or an array with one element
 954          */
 955         function closeRenameAndReopen( $newname ) {
 956         }
 957
 958         /**
 959          * Close the old file, and move it to a specified name.
 960          * Use this for the last piece of a file written out
 961          * at specified checkpoints (e.g. every n hours).
 962          * @param $newname mixed File name. May be a string or an array with one element
 963          * @param bool $open If true, a new file with the old filename will be opened
 964          *   again for writing (default: false)
 965          */
 966         function closeAndRename( $newname, $open = false ) {
 967         }
 968
 969         /**
 970          * Returns the name of the file or files which are
 971          * being written to, if there are any.
 972          * @return null
 973          */
 974         function getFilenames() {
 975                 return null;
 976         }
 977 }
 978
 979 /**
 980  * Stream outputter to send data to a file.
 981  * @ingroup Dump
 982  */
 983 class DumpFileOutput extends DumpOutput {
 984         protected $handle = false, $filename;
 985
 986         /**
 987          * @param $file
 988          */
 989         function __construct( $file ) {
 990                 $this->handle = fopen( $file, "wt" );
 991                 $this->filename = $file;
 992         }
 993
 994         /**
 995          * @param $string string
 996          */
 997         function writeCloseStream( $string ) {
 998                 parent::writeCloseStream( $string );
 999                 if ( $this->handle ) {
1000                         fclose( $this->handle );
1001                         $this->handle = false;
1002                 }
1003         }
1004
1005         /**
1006          * @param $string string
1007          */
1008         function write( $string ) {
1009                 fputs( $this->handle, $string );
1010         }
1011
1012         /**
1013          * @param $newname
1014          */
1015         function closeRenameAndReopen( $newname ) {
1016                 $this->closeAndRename( $newname, true );
1017         }
1018
1019         /**
1020          * @param $newname
1021          * @throws MWException
1022          */
1023         function renameOrException( $newname ) {
1024                         if ( !rename( $this->filename, $newname ) ) {
1025                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1026                         }
1027         }
1028
1029         /**
1030          * @param $newname array
1031          * @return mixed
1032          * @throws MWException
1033          */
1034         function checkRenameArgCount( $newname ) {
1035                 if ( is_array( $newname ) ) {
1036                         if ( count( $newname ) > 1 ) {
1037                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1038                         } else {
1039                                 $newname = $newname[0];
1040                         }
1041                 }
1042                 return $newname;
1043         }
1044
1045         /**
1046          * @param $newname mixed
1047          * @param $open bool
1048          */
1049         function closeAndRename( $newname, $open = false ) {
1050                 $newname = $this->checkRenameArgCount( $newname );
1051                 if ( $newname ) {
1052                         if ( $this->handle ) {
1053                                 fclose( $this->handle );
1054                                 $this->handle = false;
1055                         }
1056                         $this->renameOrException( $newname );
1057                         if ( $open ) {
1058                                 $this->handle = fopen( $this->filename, "wt" );
1059                         }
1060                 }
1061         }
1062
1063         /**
1064          * @return string|null
1065          */
1066         function getFilenames() {
1067                 return $this->filename;
1068         }
1069 }
1070
1071 /**
1072  * Stream outputter to send data to a file via some filter program.
1073  * Even if compression is available in a library, using a separate
1074  * program can allow us to make use of a multi-processor system.
1075  * @ingroup Dump
1076  */
1077 class DumpPipeOutput extends DumpFileOutput {
1078         protected $command, $filename;
1079         protected $procOpenResource = false;
1080
1081         /**
1082          * @param $command
1083          * @param $file null
1084          */
1085         function __construct( $command, $file = null ) {
1086                 if ( !is_null( $file ) ) {
1087                         $command .= " > " . wfEscapeShellArg( $file );
1088                 }
1089
1090                 $this->startCommand( $command );
1091                 $this->command = $command;
1092                 $this->filename = $file;
1093         }
1094
1095         /**
1096          * @param $string string
1097          */
1098         function writeCloseStream( $string ) {
1099                 parent::writeCloseStream( $string );
1100                 if ( $this->procOpenResource ) {
1101                         proc_close( $this->procOpenResource );
1102                         $this->procOpenResource = false;
1103                 }
1104         }
1105
1106         /**
1107          * @param $command
1108          */
1109         function startCommand( $command ) {
1110                 $spec = array(
1111                         0 => array( "pipe", "r" ),
1112                 );
1113                 $pipes = array();
1114                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1115                 $this->handle = $pipes[0];
1116         }
1117
1118         /**
1119          * @param mixed $newname
1120          */
1121         function closeRenameAndReopen( $newname ) {
1122                 $this->closeAndRename( $newname, true );
1123         }
1124
1125         /**
1126          * @param $newname mixed
1127          * @param $open bool
1128          */
1129         function closeAndRename( $newname, $open = false ) {
1130                 $newname = $this->checkRenameArgCount( $newname );
1131                 if ( $newname ) {
1132                         if ( $this->handle ) {
1133                                 fclose( $this->handle );
1134                                 $this->handle = false;
1135                         }
1136                         if ( $this->procOpenResource ) {
1137                                 proc_close( $this->procOpenResource );
1138                                 $this->procOpenResource = false;
1139                         }
1140                         $this->renameOrException( $newname );
1141                         if ( $open ) {
1142                                 $command = $this->command;
1143                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1144                                 $this->startCommand( $command );
1145                         }
1146                 }
1147         }
1148
1149 }
1150
1151 /**
1152  * Sends dump output via the gzip compressor.
1153  * @ingroup Dump
1154  */
1155 class DumpGZipOutput extends DumpPipeOutput {
1156
1157         /**
1158          * @param $file string
1159          */
1160         function __construct( $file ) {
1161                 parent::__construct( "gzip", $file );
1162         }
1163 }
1164
1165 /**
1166  * Sends dump output via the bgzip2 compressor.
1167  * @ingroup Dump
1168  */
1169 class DumpBZip2Output extends DumpPipeOutput {
1170
1171         /**
1172          * @param $file string
1173          */
1174         function __construct( $file ) {
1175                 parent::__construct( "bzip2", $file );
1176         }
1177 }
1178
1179 /**
1180  * Sends dump output via the p7zip compressor.
1181  * @ingroup Dump
1182  */
1183 class Dump7ZipOutput extends DumpPipeOutput {
1184
1185         /**
1186          * @param $file string
1187          */
1188         function __construct( $file ) {
1189                 $command = $this->setup7zCommand( $file );
1190                 parent::__construct( $command );
1191                 $this->filename = $file;
1192         }
1193
1194         /**
1195          * @param $file string
1196          * @return string
1197          */
1198         function setup7zCommand( $file ) {
1199                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1200                 // Suppress annoying useless crap from p7zip
1201                 // Unfortunately this could suppress real error messages too
1202                 $command .= ' >' . wfGetNull() . ' 2>&1';
1203                 return $command;
1204         }
1205
1206         /**
1207          * @param $newname string
1208          * @param $open bool
1209          */
1210         function closeAndRename( $newname, $open = false ) {
1211                 $newname = $this->checkRenameArgCount( $newname );
1212                 if ( $newname ) {
1213                         fclose( $this->handle );
1214                         proc_close( $this->procOpenResource );
1215                         $this->renameOrException( $newname );
1216                         if ( $open ) {
1217                                 $command = $this->setup7zCommand( $this->filename );
1218                                 $this->startCommand( $command );
1219                         }
1220                 }
1221         }
1222 }
1223
1224 /**
1225  * Dump output filter class.
1226  * This just does output filtering and streaming; XML formatting is done
1227  * higher up, so be careful in what you do.
1228  * @ingroup Dump
1229  */
1230 class DumpFilter {
1231
1232         /**
1233          * @var DumpOutput
1234          * FIXME will need to be made protected whenever legacy code
1235          * is updated.
1236          */
1237         public $sink;
1238
1239         /**
1240          * @var bool
1241          */
1242         protected $sendingThisPage;
1243
1244         /**
1245          * @param $sink DumpOutput
1246          */
1247         function __construct( &$sink ) {
1248                 $this->sink =& $sink;
1249         }
1250
1251         /**
1252          * @param $string string
1253          */
1254         function writeOpenStream( $string ) {
1255                 $this->sink->writeOpenStream( $string );
1256         }
1257
1258         /**
1259          * @param $string string
1260          */
1261         function writeCloseStream( $string ) {
1262                 $this->sink->writeCloseStream( $string );
1263         }
1264
1265         /**
1266          * @param $page
1267          * @param $string string
1268          */
1269         function writeOpenPage( $page, $string ) {
1270                 $this->sendingThisPage = $this->pass( $page, $string );
1271                 if ( $this->sendingThisPage ) {
1272                         $this->sink->writeOpenPage( $page, $string );
1273                 }
1274         }
1275
1276         /**
1277          * @param $string string
1278          */
1279         function writeClosePage( $string ) {
1280                 if ( $this->sendingThisPage ) {
1281                         $this->sink->writeClosePage( $string );
1282                         $this->sendingThisPage = false;
1283                 }
1284         }
1285
1286         /**
1287          * @param $rev
1288          * @param $string string
1289          */
1290         function writeRevision( $rev, $string ) {
1291                 if ( $this->sendingThisPage ) {
1292                         $this->sink->writeRevision( $rev, $string );
1293                 }
1294         }
1295
1296         /**
1297          * @param $rev
1298          * @param $string string
1299          */
1300         function writeLogItem( $rev, $string ) {
1301                 $this->sink->writeRevision( $rev, $string );
1302         }
1303
1304         /**
1305          * @param $newname string
1306          */
1307         function closeRenameAndReopen( $newname ) {
1308                 $this->sink->closeRenameAndReopen( $newname );
1309         }
1310
1311         /**
1312          * @param $newname string
1313          * @param $open bool
1314          */
1315         function closeAndRename( $newname, $open = false ) {
1316                 $this->sink->closeAndRename( $newname, $open );
1317         }
1318
1319         /**
1320          * @return array
1321          */
1322         function getFilenames() {
1323                 return $this->sink->getFilenames();
1324         }
1325
1326         /**
1327          * Override for page-based filter types.
1328          * @param $page
1329          * @return bool
1330          */
1331         function pass( $page ) {
1332                 return true;
1333         }
1334 }
1335
1336 /**
1337  * Simple dump output filter to exclude all talk pages.
1338  * @ingroup Dump
1339  */
1340 class DumpNotalkFilter extends DumpFilter {
1341
1342         /**
1343          * @param $page
1344          * @return bool
1345          */
1346         function pass( $page ) {
1347                 return !MWNamespace::isTalk( $page->page_namespace );
1348         }
1349 }
1350
1351 /**
1352  * Dump output filter to include or exclude pages in a given set of namespaces.
1353  * @ingroup Dump
1354  */
1355 class DumpNamespaceFilter extends DumpFilter {
1356         var $invert = false;
1357         var $namespaces = array();
1358
1359         /**
1360          * @param $sink DumpOutput
1361          * @param $param
1362          * @throws MWException
1363          */
1364         function __construct( &$sink, $param ) {
1365                 parent::__construct( $sink );
1366
1367                 $constants = array(
1368                         "NS_MAIN"           => NS_MAIN,
1369                         "NS_TALK"           => NS_TALK,
1370                         "NS_USER"           => NS_USER,
1371                         "NS_USER_TALK"      => NS_USER_TALK,
1372                         "NS_PROJECT"        => NS_PROJECT,
1373                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1374                         "NS_FILE"           => NS_FILE,
1375                         "NS_FILE_TALK"      => NS_FILE_TALK,
1376                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1377                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1378                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1379                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1380                         "NS_TEMPLATE"       => NS_TEMPLATE,
1381                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1382                         "NS_HELP"           => NS_HELP,
1383                         "NS_HELP_TALK"      => NS_HELP_TALK,
1384                         "NS_CATEGORY"       => NS_CATEGORY,
1385                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1386
1387                 if ( $param { 0 } == '!' ) {
1388                         $this->invert = true;
1389                         $param = substr( $param, 1 );
1390                 }
1391
1392                 foreach ( explode( ',', $param ) as $key ) {
1393                         $key = trim( $key );
1394                         if ( isset( $constants[$key] ) ) {
1395                                 $ns = $constants[$key];
1396                                 $this->namespaces[$ns] = true;
1397                         } elseif ( is_numeric( $key ) ) {
1398                                 $ns = intval( $key );
1399                                 $this->namespaces[$ns] = true;
1400                         } else {
1401                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1402                         }
1403                 }
1404         }
1405
1406         /**
1407          * @param $page
1408          * @return bool
1409          */
1410         function pass( $page ) {
1411                 $match = isset( $this->namespaces[$page->page_namespace] );
1412                 return $this->invert xor $match;
1413         }
1414 }
1415
1416 /**
1417  * Dump output filter to include only the last revision in each page sequence.
1418  * @ingroup Dump
1419  */
1420 class DumpLatestFilter extends DumpFilter {
1421         var $page, $pageString, $rev, $revString;
1422
1423         /**
1424          * @param $page
1425          * @param $string string
1426          */
1427         function writeOpenPage( $page, $string ) {
1428                 $this->page = $page;
1429                 $this->pageString = $string;
1430         }
1431
1432         /**
1433          * @param $string string
1434          */
1435         function writeClosePage( $string ) {
1436                 if ( $this->rev ) {
1437                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1438                         $this->sink->writeRevision( $this->rev, $this->revString );
1439                         $this->sink->writeClosePage( $string );
1440                 }
1441                 $this->rev = null;
1442                 $this->revString = null;
1443                 $this->page = null;
1444                 $this->pageString = null;
1445         }
1446
1447         /**
1448          * @param $rev
1449          * @param $string string
1450          */
1451         function writeRevision( $rev, $string ) {
1452                 if ( $rev->rev_id == $this->page->page_latest ) {
1453                         $this->rev = $rev;
1454                         $this->revString = $string;
1455                 }
1456         }
1457 }
1458
1459 /**
1460  * Base class for output stream; prints to stdout or buffer or wherever.
1461  * @ingroup Dump
1462  */
1463 class DumpMultiWriter {
1464
1465         /**
1466          * @param $sinks
1467          */
1468         function __construct( $sinks ) {
1469                 $this->sinks = $sinks;
1470                 $this->count = count( $sinks );
1471         }
1472
1473         /**
1474          * @param $string string
1475          */
1476         function writeOpenStream( $string ) {
1477                 for ( $i = 0; $i < $this->count; $i++ ) {
1478                         $this->sinks[$i]->writeOpenStream( $string );
1479                 }
1480         }
1481
1482         /**
1483          * @param $string string
1484          */
1485         function writeCloseStream( $string ) {
1486                 for ( $i = 0; $i < $this->count; $i++ ) {
1487                         $this->sinks[$i]->writeCloseStream( $string );
1488                 }
1489         }
1490
1491         /**
1492          * @param $page
1493          * @param $string string
1494          */
1495         function writeOpenPage( $page, $string ) {
1496                 for ( $i = 0; $i < $this->count; $i++ ) {
1497                         $this->sinks[$i]->writeOpenPage( $page, $string );
1498                 }
1499         }
1500
1501         /**
1502          * @param $string
1503          */
1504         function writeClosePage( $string ) {
1505                 for ( $i = 0; $i < $this->count; $i++ ) {
1506                         $this->sinks[$i]->writeClosePage( $string );
1507                 }
1508         }
1509
1510         /**
1511          * @param $rev
1512          * @param $string
1513          */
1514         function writeRevision( $rev, $string ) {
1515                 for ( $i = 0; $i < $this->count; $i++ ) {
1516                         $this->sinks[$i]->writeRevision( $rev, $string );
1517                 }
1518         }
1519
1520         /**
1521          * @param $newnames
1522          */
1523         function closeRenameAndReopen( $newnames ) {
1524                 $this->closeAndRename( $newnames, true );
1525         }
1526
1527         /**
1528          * @param $newnames array
1529          * @param bool $open
1530          */
1531         function closeAndRename( $newnames, $open = false ) {
1532                 for ( $i = 0; $i < $this->count; $i++ ) {
1533                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1534                 }
1535         }
1536
1537         /**
1538          * @return array
1539          */
1540         function getFilenames() {
1541                 $filenames = array();
1542                 for ( $i = 0; $i < $this->count; $i++ ) {
1543                         $filenames[] = $this->sinks[$i]->getFilenames();
1544                 }
1545                 return $filenames;
1546         }
1547
1548 }
1549
1550 /**
1551  * @param $string string
1552  * @return string
1553  */
1554 function xmlsafe( $string ) {
1555         wfProfileIn( __FUNCTION__ );
1556
1557         /**
1558          * The page may contain old data which has not been properly normalized.
1559          * Invalid UTF-8 sequences or forbidden control characters will make our
1560          * XML output invalid, so be sure to strip them out.
1561          */
1562         $string = UtfNormal::cleanUp( $string );
1563
1564         $string = htmlspecialchars( $string );
1565         wfProfileOut( __FUNCTION__ );
1566         return $string;
1567 }